CFD Online Discussion Forums - the periodic time for unsteady cases

A first approximation would be to estimate the Strouhal Number (a cylinder is generally assumed to be around .2 depending on your Reynold's number). From there you can calculate the vortex shedding frequency and according to the Fluent manual, you're going to want about 20 iterations per shedding cycle. If you can't find the Strouhal number for your geometry, then try to find the Strouhal number for the closest shape and use a smaller time step than would be needed. For example, if you're geometry is somewhat cylindrical, then you can assume a St of .2... if that gives you a time step size of .02 seconds, then maybe you run with .01 seconds. If your model keeps converging within a few iterations, then you can up the time step. If the model doesn't converge within the 30 time steps, then you should try an even smaller time step. You want the model to converge between 10 and 30 iterations per time step. Really small time steps shouldn't be a surprise depending on your model.

Hope this helps, and good luck, Jason